An image forming apparatus includes upstream-side and downstream-side detection parts and a control section. The detection parts are disposed on a moving path of an intermediate transfer belt or a conveyor belt and detect a speed of the belt. The control section performs control to correct color misregistration of images composed of colors to be formed on the intermediate transfer belt or a recording medium conveyed by the conveyance belt on the basis of the detection result. The control unit calculates a difference between the speed detected by the upstream-side detection part and the speed detected by the downstream-side detection part after a predetermined time elapses since the upstream-side detection part detects the speed and performs the control on the basis of the difference. The predetermined time is obtained by dividing a distance between the detection parts by a target speed of the belt.